Watershed image segmentation based on nonlinear combination morphology filter

Traditional watershed algorithm´ ability to inhibit noise is not that strong, so causing regional minima and leading to over-segmentation. So a watershed image segmentation algorithm based on the nonlinear combination morphology filter has been put forward. First of all, we define the nonlinear combination morphology filter with opening-closing operators and closing-opening operators for image filtering; Secondly, we design a new morphology watershed algorithm with inner and external marks, and also define the regional minima to inner marks from the low frequency components of the gradients and external marks between the region, the inner and external marks changes along with the image information, thus has realized the adaptive image segmentation. Simulation results show that the new algorithm can reduce over-segmentation arising from false local minima in a gradient image which is caused by the noise, which could accurately realize the image segmentation.